Add warning for large datasets on CPU and reference to tabpfn-client API

[Krishnadubey1008]

This PR adresses issue #192

Changes

• Added a check in the fit method to determine if the code is running on CPU and if the dataset size exceeds 1000 samples.
• Displayed a warning message to inform users of potential performance issues and suggested using the tabpfn-client API or a GPU.

[CLAassistant]

All committers have signed the CLA.

[noahho]

Hi! Please sign the CLA to make your PR useable in our package! :-)

Pull Request Overview

This PR adds a runtime warning for users when running the fit method on a CPU with a large dataset, suggesting the use of a GPU or the tabpfn-client API to improve performance.

• Added a conditional check for CPU usage and dataset size in the fit method.
• Displays a warning message with a reference to the tabpfn-client API.

Comments suppressed due to low confidence (2)

src/tabpfn/regressor.py:417

• Consider adding a 'stacklevel' argument to warnings.warn (e.g., stacklevel=2) so that the warning message points to the caller's line, improving traceability for the user.

warnings.warn(

src/tabpfn/regressor.py:416

• Please add corresponding unit tests to validate the warning behavior when running on CPU with a large dataset.

if self.device == 'cpu' and X.shape[0] > 1000:

[Krishnadubey1008]

@noahho earlier I had unverified email connected to git commits now I have fixed it and also signed CLA. Please Review the PR

[Krishnadubey1008]

@noahho Please tell me according to the copilot overview should i need to add unit tests(at test_regressor_interface.py) to validate the warning behavior when running on CPU with a large dataset.

def test_cpu_large_dataset_runtime_warning():
    model = TabPFNRegressor(device='cpu')
    X_large = np.random.rand(1001, 10)
    y_large = np.random.rand(1001)

    with pytest.warns(RuntimeWarning, match="Running on CPU with large dataset"):
        model.fit(X_large, y_large)